County vs. State

Hi Ernesto, This actually is a pretty common source of confusion. In general a larger region has less leakages due to imports and therefore is typically a larger impact. However, it is actually quite possible for an industry to be more concentrated in terms of output per worker, and compensation per worker in a region where there is specialization than in the state as a whole, or where the remaining state is more rural that the county where the economic change is occurring. This is one of the reasons that we developed the ability in Version 3.0 to conduct multi-regional analyses. You can see this by comparing the Events between the two models (county and state). There are built in tool tips in the software, that allow you to see these output per worker and compensation per worker values by hovering the mouse over the employment and compensation fields. You will likely see that the compensation per worker and output per worker are larger in your county. Since it sounds like you might not have the data to do MRIO, there is a way that you can approximate an MRIO. Basically the goal is to make the State Model's data for the Directly affected industry(s) match the county’s. 1. In the county model, navigate to the Customize Study Area Data screen. Next select the first Sector that you need to modify from the side bar at the left-hand side of the screen. You will want to write down all the information in either the Per Worker column or the totals column. Repeat this for any other Direct Industries that you are comparing between the county and the state. 2. Open the State Model and navigate to the Customize Study Area Data screen. Select your industries as before and enter the recorded data from the county file into the same state’s column (per worker or totals) so that the industry data for the State Model is identical to the county model. When you update the sheet, the numbers you entered will be balanced in the column you did not change. 3. Save your changes. Repeat this for any other Direct Industries that you are comparing. 4. Once all industries have been modified reconstruct the Model through the Options menu. 5. Once the model is rebuilt you can run your Activities and Events in both models. You can then subtract county from the state data considering the leakages to be the differences between the amount of available inputs to intermediate and final demand. Please let us know if you have any additional questions.

Thank you. This seem pretty straight forward; Just a couple of clarifications: [i]1. In the county model, navigate to the Customize Study Area Data screen. Next select the first Sector that you need to modify from the side bar at the left-hand side of the screen. You will want to write down all the information in either the Per Worker column or the totals column. Repeat this for any other Direct Industries that you are comparing between the county and the state.[/i] When you say "Direct Industries that you are comparing" you mean all the industries where I am entering Activities and Events, correct? [i]5. Once the model is rebuilt you can run your Activities and Events in both models. You can then subtract county from the state data considering the leakages to be the differences between the amount of available inputs to intermediate and final demand.[/i] Do I need to report the difference? or can I just state that whatever results I get from the state analysis is the impact on the state and whatever I get from the County is the impact on the county? Also, I have one more questions. Our client also wants us to do the analysis for the City. We have the zip code data and have created a model with all the zip codes within the City. How would I run this so that it is consistent with the results for the County and the State as described above? Thanks again. Regards.

Hi Ernesto, Sorry that wasn't completely clear. This is a set of instructions we use whenever this type of question is asked. If you are only creating an Event for one Sector, you will only need to write down the information for that Sector. You will need to get information for each Sector for which you create an Event line. In regards to number 5 that is up to you. You will just want to check and confirm that both models have IDENTICAL Direct impacts. Since the ZIP Code level relationships should match the county relationships, you should be able to just enter your initial values as you did in the county level model and see the same Direct Output at the ZIP Code level as you have in the county. As long as your Direct Effects match those at the county level the differences in the total will reflect the leakages of commodity and service purchases and commuting. Please let us know if you have any additional questions.

Thanks that's clear. Just one more question: what if one of the changes I am modeling is a household income? change?

Ok, I followed your instructions and updated the State model to reflect the County. Since this is a construction analysis, I modified Industries 34, 36, and 37 for Connecticut with the per worker numbers from the county. See MS Word Attachment. Then I ran my analysis for the City, the County, and the State (each independently). the results for the City and County make sense. However, there are two issues with the State results: 1) The direct impacts are not identical eventhough I have entered the same activities/events 2) The induced jobs at the state level are lower than the county. See excel file. Any ideas why this occurs? I can see why #2 above occurs if the state has lower income levels than the county, but it would still be difficult to explain to policy makers why this is the case. Thank you

Hi Ernesto, It looks like something might have gone wrong in the Edit screen based on the information that your have provided. Can you please email your state of CT model to implangroup@implan.com. Thanks!

How do I do that?

do you mean the 'impdb' file?

I sent you the file via dropbox link. Did you get it?

Yes we did, we are working on it right now and should get back to your shortly.

It looks like you forgot to change sector 34's employment total in the Customize>Study Area Data screen. We were able to change the State model you sent to match the county's ratios. Please let us know if you have any further questions.

I wasn't aware that I was supposed to change 'Total Employment', I had only changed the 'Per Worker' column. Just to be sure, in the state model, I need to change all of the numbers that are highlighted with a red box in the attachment? [attachment:1]County factors.docx[/attachment]

We apologize we didn't make that clear before. As regards your attachment, it doesn't appear that it made it through, if you would like us to review it, please send it to the email address listed above. You will need to make this change because as it is now the per-worker values match, but the employment total did not. Hence you have changed the level of production per worker to match your county but not you your workers, so you will see that while the per worker relationships match, the totals do not, but we do need them to match as well. The relationships to used generate the industry total will update to match the county, when you change the employment total. When these relationships are identical for both the total and the per worker values, your Direct effects should match in between the state and county model. The remaining differences will specifically arise since the state is a larger region and the makeup of state-wide industries are more rounded than regional makeups of industries. Please let us know if after making this change, you are still having problem getting the Direct effects to match up.

Thanks again. I updated the "Total Employment" cell along with the "per worker" column and voila! it's solved. Thanks. However, while direct impacts now match for both the state and the county, and indirect impacts are large at the state level than the county (what you would expect), the indirect impacts are larger at the county level than at the state level? that doesn't make sense. Any ideas how to solve this? A separate question (although it may be related to the question above): So I modified the impacted industries at the state level to mirror the county. However, one of the activities I a am modeling is an increase in household income? how would I modify state model so that I can analyze a household income activity both at the state level and county level? Thank you.

Hi Ernesto, We're glad we got the one problem resolved. We have played around a little with the issue of the reduced Induced impact at the state, and we are seeing and being able to duplicate what you are seeing. We are looking into some reason's why that might be and will post something later this morning in regards to what we find. Thank you for your patience.

Thanks. Please keep me posted. We have to submit our report tomorrow, so it'd would be great to sort this out ASAP. Regards.

Hi Ernesto, It took us a little longer than we had hoped to check this out, but it actually is a real phenomena in Connecticut. One consideration is that even with the edits there still are some small discrepancies based on rounding in the value of the employment compensation and the value of proprietor income. If you overwrite the Event compensation and income in the state to exactly match the county values that will cause the discrepancy in Employment to close some and will solve the lower Induced Output effect. As regards the overall Employment effect. The workers at the state level for the Indirect industries are apparently paid somewhat less than the workers in New Haven, which is what is causing the Induced Employment and thus the overall Employment to be smaller. A couple of additional notes: [ul] [li]You don't need to make any changes for the Household Income Change, those are directly comparable results between the county and state because there are no "Direct" relationships here that need to be captured[/li] [li]You can test running a Labor Income Change in both the state and the county to duplicate the test of the reduced Induced Employment [/li] [li]While typically state impacts are larger for Direct and Induced there are circumstances where state ratios of income per worker or other factors may be significantly lower than a county's. This appears to be the case in this region. These are real constraints in the reported values for the state.[/li] [/ul] Please let us know if you have any additional questions and again we apologize that it took us so long to diagnose exactly what was going on.

Thank you. Based on your response below, I did the following: In the state model, in my list of events, I went and filled in the industry sales. IMPLAN automatically fills in the Employment, Employment Compensation, and Proprietor Income. Since, I had already modified the affected industries in the state model, the values that implan calculates in the 'events' window for Employment, Employee Compensation, and Proprietor Income were very close to the ones in the County model. In fact the employment figures were exactly the same. The values for Employee Compensation and Proprietor Income were very close but not exactly the same. So I changed those values to match the ones in the county. Then I ran the model again. However, nothing changed! You write: "The workers at the state level for the Indirect industries are apparently paid somewhat less than the workers in New Haven, which is what is causing the Induced Employment and thus the overall Employment to be smaller...While typically state impacts are larger for Direct and Induced there are circumstances where state ratios of income per worker or other factors may be significantly lower than a county's. This appears to be the case in this region. These are real constraints in the reported values for the state." Does this means that there is nothing I can do about the fact that the estimated induced effects for employment at the state level are lower than the county? That'd be hard to explain to policy makers. Could I ignore the estimated induced employment and labor income impacts at the state level and replace them with the county results? This would imply that the effects at the state level are at least as large as the effects on the county.

So I looked around the forum and saw a response to a similar question to mine. it looks like there may be a way to address the issue of the state having a lower induce impact than the county, namely: "1. The preferred methodology if you have the state package available to you, is the multiregional methodology (MRIO). Where you build your county model and a rest of state model that includes all the counties except the one you are examining the petrochemical industry in. You would then link the rest of state model to your petrochemical Scenario to get the results. Here are some links on setting up and running MRIO's. implan.com/V4/index.php?option=com_multi...alysis&Itemid=71" Do you think this approach would solve the problem I have? I bought the IMPLAN data for the County of New Haven (County Plus, since I also need to look at the impacts in a county subarea) and the State of Connecticut. I guess I should have bought the dataset that has all of the counties in the state of Connecticut instead of just the overall state. If the approach described above will resolve my problem, then I'll buy the state package that has all the counties, but can I get a credit for the state package I already bought? Thank you for your assistance.

Hi Ernesto, You are correct, the MRIO is the best solution to this problem and if you have the ability/budget to get the state package that is the best way to resolve what you are seeing. MRIO shows you what happens in the county and what happens in the rest of the state as a result of the counties activity, so since the results are not tied together like they are in the estimation method you have been using, you will be able to see what the county does and how that impacts the state. You are correct that there is no way to change the lower induced Employment this a real reflection of the fact that on average across various industries in the state, a lower wage is paid to workers than in the county of New Haven. As regards a discount, you can get credit for the state total file you purchased towards the purchase of the CT state package, but with a county plus package, you are purchasing the ZIP Code data that does not have overlapping coverage with the state package, so unfortunately that one will not count towards the purchase of the state package. Please call us at 651.439.4421 and we can assist you with this order, and please let us know if you have any additional questions or concerns.